1. Field of the Invention
The present invention relates to an image pickup system, an image processing method, and a computer program product for extracting an edge component with respect to a signal from an image pickup device.
2. Description of the Related Art
A processing for extracting an edge component with respect to a signal from an image pickup device is roughly divided into a single filter processing such as Laplacian and a directional plural-filter processing. As a latter example, as in Japanese Examined Patent Application Publication No. 6-90724, for example, there is disclosed a technology of performing a directional unsharp mask processing by selecting a rectangular region in which a difference with respect to a smoothed signal is maximized among four-direction rectangular regions. With this configuration, an influence of a dotted defect such as dust is suppressed, and a higher precision edge component can be extracted. In addition, Japanese Patent No. 3289435 discloses a technology of calculating a weighting factor from a step edge, an impulse signal, a noise, or the like to correct an edge signal on the basis of this factor. With this configuration, the edge signal is shaped, and it is possible to extract a higher precision edge component. Furthermore, Japanese Unexamined Patent Application Publication No. 2000-306089 discloses a technology of frequency-decomposing a signal through wavelet, Laplacian pyramid, or the like and correcting a high frequency component from vector information of a low frequency component. With this configuration, the edge signal is shaped, and also a noise component is suppressed, thus making it possible to obtain a high precision signal.
According to Japanese Examined Patent Application Publication No. 6-90724, it is possible to identify a large dotted noise with a large luminance difference such as the dust. However, it is not possible to deal with an identification of an isotropically generated noise such as a random noise due to the image pickup device or a slight noise with small luminance difference, and there is a problem that it is not possible to stably extract an edge component from various noises. In addition, according to Japanese Patent No. 3289435, in a case where the noises are large, it is possible to suppress the influence of the noises by generating a small weighting factor. However, as the noise component itself remains, it is not possible to deal with an execution of a strong edge emphasis, and there is a problem that a degree of freedom for image creation is small. Furthermore, according to Japanese Unexamined Patent Application Publication No. 2000-306089, the vector information for the correction is obtained from the low frequency signal. A resolution of the vector information depends on a decomposition level of the wavelet or the Laplacian pyramid, but the resolution becomes ½ when the decomposition is performed once, and the resolution becomes ¼ when decomposition is performed twice. For this reason, it is not possible to deal with the shaping of the edge signal at the resolution of the original signal, and there is a problem that a high precision edge component in a pixel unit cannot be generated.
The present invention has been made in view of the above-mentioned circumstances, and it is an object to provide an image pickup system, an image processing method, and a computer program product with which it is possible to extract a high precision edge component with respect to a signal where a noise is mixed as well by combining a noise reduction processing with an edge extraction processing.